computational analysis and proposed redesign of the 3-stage axial compressor … · 2017. 5....

Computational Analysis and Proposed Redesign of the

3-Stage Axial Compressor in the 11-by-11-Foot

Transonic Wind Tunnel at NASA Ames Research Center

Sameer Kulkarni

Joe Veres

Phil Jorgenson

Tim Beach

NASA Glenn Research Center

5/24/2017

1

Contents

• Background and Motivation

• Features of the Existing Compressor and Design Constraints

• Approach

• Validating Tools Against Existing Data

• Recommendations for a New Design

2

Background on Ames 11x11 TWT

• 11- by 11-Foot Transonic Wind Tunnel (11-Foot

TWT) Facility at NASA Ames Research Center in

Moffett Field, California “has been instrumental in

the development of virtually every domestically

produced commercial transport and military fixed-

wing airframe since the 1960s. The facility is used

extensively for airframe testing and aerodynamic

studies and has played a vital role in every manned

space flight program.” 1

• Closed-loop tunnel with Mach range 0.20 to 1.40,

stagnation pressure range 3.0 to 32.0 psia, Rn

range 0.30 to 9.6 million per foot.

• Airflow provided by 24 foot diameter, 3-stage axial

compressor powered by variable-speed induction

motors.

3

1 https://www.nasa.gov/centers/ames/orgs/aeronautics/windtunnels/11x11-wind-tunnel.html

Unitary Plan Wind Tunnel

4

Image from Baals, Donald D., “Wind Tunnels of NASA,” NASA SP-440, 1981.

Motivation

• Compressor was designed ~1950s. Aluminum blades were used to minimize

weight due to rotordynamics concerns. These blades are sensitive to

damage/erosion.

• Blades are inspected for impacts/damage every 50 hours and overhauled every

2400 hours, wherein the blades are removed for sanding and penetrant

inspection.

– Overhaul is approx. 1 man-year effort and 1 month of facility downtime.

• Re-blading with new hollow steel blades was planned to stretch inspections to

200 hours and obviate blade overhauls.

• Several facility upgrades over 60+ years (flow conditioning devices,

instrumentation) have increased tunnel blockage and reduced peak test section

Mach number capability.

Opportunity to increase test section Mach number capability to 1.45+ by

increasing compressor pressure ratio via new blade design.

5

Compressor Case Split During Overhaul

6

Spare Rotor Blades and Blades Awaiting Reburb

7

Spare Rotor Blades and Blades Awaiting Reburb

8

GRC Team In-between Rotor Disks

9

Features of the Existing Compressor

10

• 3-stage axial compressor with inlet guide vanes (IGVs) and exit guide vane (EGVs).

• 54 IGVs, 52 rotor blades per stage, 34 stator 1&2 vanes, 58 stator 3 vanes, 60 EGVs.

• IGVs are variable camber (hinged at mid-chord); IGV flap can vary from -7.5° (less pre-

swirl) to +19.5° (more pre-swirl). Nominal 0° position gives +33° pre-swirl into rotor 1.

• Hub and casing flow paths have constant radius of 8.5 ft and 12 ft respectively.

• Approx. 0.5 inch rotor tip clearance.

• Rotational speed ranges from 150-650 RPM (rotor tip speed 190-815 ft/s).

• Test section Mach number and Reynolds number are set by varying compressor speed,

IGV camber, and compressor inlet pressure.

• At peak test section Mach number of ~1.45 (empty test section), compressor inlet

corrected flow rate is ~7000 lb/s and compressor overall pressure ratio is ~1.4.

Cross-sectional View of the Compressor Case

11

Design Constraints

• Rotor blade shapes may change, but blade count may not (re-use existing disks).

• Desire to keep new rotor shape identical across each stage to minimize

tooling/fabrication costs.

• Stator vane shapes and vane counts may not change, but re-staggering existing

stators is possible.

• IGVs must be remain unchanged due to complexities in removing/replacing

actuators. May be possible to close IGV flap additional 2.5° for -10° total closure

from nominal.

• Available margins in motor, shaft, and bearings should allow for increasing

maximum rotational speed from 650 to 695 RPM.

12

Operating lines for the existing compressor (left) and facility (right). Red lines indicate target for a new design.

Approach

• Model the performance of the existing compressor and

compare to test data to calibrate numerical tools:

– Turbomachinery design/analysis code HT03002.

– Turbomachinery RANS 3D CFD solvers APNASA3 and ADPAC4.

• Iterate on new compressor design (within identified design

constraints) using HT0300 to achieve at least a 10%

increase in total pressure rise.

• Validate predicted performance of new design with CFD

using APNASA.

13

2 Hearsey, R. M., “Program HT0300 NASA 1994 Version,” Doc. No. D6-81569TN, Volumes 1 and 2, The Boeing Company, 1994.3 Adamczyk, J. J., "Model Equation for Simulating Flows in Multistage Turbomachinery," NASA TM-86869, 1985.4 Hall, E. J., Heidegger, N. J. and Delaney, R. A., "ADPAC v I .O - User's Manual," NASA CR-1999-206600, 1999.

Validating Tools Against Existing Data

• Existing compressor geometry with nominal IGV setting was used as input for HT0300 in

analysis mode to generate performance predictions at 634 RPMC.

• Geometry definition was then used to generate meshes for CFD analysis using APNASA

(905x51x51 grid). Initial APNASA simulation is referenced herein as “APNASA A”

• These two results were compared to data collected during a 1997 facility checkout.

14

1.15

1.2

1.25

1.3

1.35

1.4

6700 6800 6900 7000 7100 7200 7300 7400 7500

Tota

l Pre

ssu

re R

atio

Compressor Inlet Corrected Flow (lbm/s)

DATA

HT0300

APNASA A

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4

Frac

tio

n S

pan

Total Pressure Ratio

DATA North Rake

DATA South Rake

APNASA A

634 RPMC speedline with nominal IGV angle.

Open symbols indicate diffusion factors

exceeding 0.5 in the design code or

unsteady/unconverged CFD results.

Radial profile of exit P0 normalized by inlet P0.

Comparison of operating points at ~6890 lb/s.







15

1.15

1.2

1.25

1.3

1.35

1.4

6700 6800 6900 7000 7100 7200 7300 7400 7500

Tota

l Pre

ssu

re R

atio


DATA

HT0300

APNASA A

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4

Frac

tio

n S

pan


DATA North Rake

DATA South Rake

APNASA A







Initial CFD has

significantly

underestimated

performance

Lower 50% of span

appears to be

separated in the

initial CFD result

Circumferentially mass-averaged axial

velocity contour of the initial CFD

result showing massive hub separation.







16

1.15

1.2

1.25

1.3

1.35

1.4

6700 6800 6900 7000 7100 7200 7300 7400 7500

Tota

l Pre

ssu

re R

atio


DATA

HT0300

APNASA A

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4

Frac

tio

n S

pan


DATA North Rake

DATA South Rake

APNASA A







Initial CFD has

significantly

underestimated

performance

Lower 50% of span

appears to be

separated in the

initial CFD result

Evidence of

separated flow in the

data, but to a lesser

spanwise extent

• Compressor hardware was examined for physical or geometrical features that were not

included in the initial CFD simulations.

Potential Sources of Performance Mismatch

• Stator “button” endwall gap leakages:

Resettable stator vanes have these gaps to

avoid interference with flowpath hardware.

• Gaps can be gridded and included in CFD

solution domain or modeled as periodic

boundary conditions.

17


18

• Under-stator cavity leakage driven by static pressure gradients at the

stator hubs – these are typically sealed via labyrinth seals but are

unsealed in this compressor.

• Leakage path can be gridded and included in the CFD solution domain

or modeled as mass flow inflow/outflow boundary conditions along the

hub.


• ADPAC was used to grid the under-stator cavities and generate a CFD

solution.

• Total mass flow recirculation in each cavity was found to be approx.

0.5% of compressor inlet mass flow rate, with reinjection angles of 25°

and 20° upstream stator 1 and 2 respectively.

19

Meridional contour of static pressure showing the under-stator cavity flow

recirculation, with absolute velocity vectors colored by axial velocity magnitude

Performance Impact of Modeling Stator Cavity and

Button Leakages

• Additional APNASA simulations generated including stator

Button gap leakages (APNASA B) and under-stator Cavity

leakages (APNASA C).

20

1.15

1.2

1.25

1.3

1.35

1.4

6700 6800 6900 7000 7100 7200 7300 7400 7500

Tota

l Pre

ssu

re R

atio


DATA

APNASA A

APNASA B

APNASA C

634 RPMC speedline with nominal IGV angle. Open symbols

indicate diffusion factors exceeding 0.5 in the design code or


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4Fr

acti

on

Sp

an


DATA North RakeDATA South RakeAPNASA AAPNASA BAPNASA C

Radial profile of exit P0 normalized by inlet P0. Comparison

of operating points at ~6890 lb/s.


Button Leakages

• APNASA B: Endwall gap leakage flow from

pressure surface re-energizes low momentum

fluid associated with corner separations on

suction surface and hub/case

• APNASA C: Corner separation at hub is

reduced due to smaller boundary layer driven by

high momentum fluid leaking into the flowpath

from the under-stator cavity upstream of the

stator leading edge.

21

Cross-passage contours of axial velocity at 70% chord of

stator 3.


Button Leakages

• Stator cavity leakage and button leakage models rolled into APNASA *

• Performance comparison has improved but it is clear that computational

stability and high loadings are problematic in the existing design.

• Stator cavity leakage and button leakage models were applied to new

design iterations to capture physics of the real compressor.

22

1.15

1.2

1.25

1.3

1.35

1.4

6700 6800 6900 7000 7100 7200 7300 7400 7500

Tota

l Pre

ssu

re R

atio


DATA

HT0300

APNASA A

APNASA *

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1 1.01

Fra

ctio

n S

pa

n

Normalized Total Pressure Ratio

DATA North Rake

DATA South Rake

APNASA *

634 RPMC speedline with nominal IGV angle. Open

symbols indicate diffusion factors exceeding 0.5 in the

design code or unsteady/unconverged CFD results.

Radial profile of exit pressure ratio, normalized by the local

maximum exit pressure ratio. APNASA * shows good

agreement with rake data in profile shape.

Diffusion Factor Limitations

• Diffusion factor for a compressor blade element is a design parameter which is

correlated with loss, separation, stability. It is defined as:

23

𝐷 = 1 −𝑉2𝑉1

+∆𝑉𝜃2𝜎𝑉1

Diffusion due to

deceleration of flow

in the relative frame

across the blade

Turning associated

with the suction

surface of the blade

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6Fr

acti

on

Sp

anDiffusion Factor

7236 lbm/s

7040 lbm/s

6838 lbm/s

– 𝜎: solidity (chord/pitch)

– V1: Axial velocity at blade row inlet

– V2: Axial velocity at blade row exit

– ∆𝑉𝜃: Difference in inlet and exit relative tangential velocities

Stator 3 diffusion factor predicted by HT0300 for

a range of operating points with nominal IGV

angle at 634 RPMC.

• Typically, losses spike for D>0.5 and flow

separation from the suction surface of the blade

leads to flow instability and compressor stall.

• Numerical tools suggest that the existing

compressor operates at D>0.5 at high speeds,

likely due to constant annular area which over-

diffuses the flow and the low solidity of stators,

especially at the hub.

Recommendations for a New Design

• Design speed was increased from 650 RPM to 690 RPM.

• Rotor blade inlet and exit angles were modified to accommodate changes in

incidence at this higher speed.

• Identical rotor blade shapes were used across all three stages.

• Existing stator 1 and 2 vanes were re-staggered to reduce incidence angles.

• New airfoil shapes for stator 3 and EGV were proposed which incorporated an

increase in hub radius from 8.5 ft to 9.4 ft.

• Stator 3 chord was increased by 15% at the case and 21.5% at the hub, resulting

in approx. 14% increased solidity averaged across the span.

24

𝐷 = 1 −𝑉2𝑉1

+∆𝑉𝜃2𝜎𝑉1

Recommendations for a New Design

• Proposed design increases overall total pressure rise by 10.5% with a

corresponding decrease in stator 3 diffusion factor of 50%.

• It is estimated that this design may increase tunnel Mach number

capability to 1.50.

25

1.2

1.22

1.24

1.26

1.28

1.3

1.32

1.34

1.36

1.38

1.4

1.42

1.44

1.46

6700 6800 6900 7000 7100 7200 7300 7400 7500 7600 7700

Tota

l Pre

ssu

re R

atio


DATA (Existing Design)

HT0300 Existing Design

HT0300 Redesign

APNASA Existing Design

APNASA Redesign

Speedlines with -7.5° IGV angle. Open symbols indicate diffusion

factors exceeding 0.5 in the design code. Existing design results are

at 634 RPMC and redesign results are at 653 RPMC.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6

Frac

tio

n S

pan

Diffusion Factor

Existing Design

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6

Frac

tio

n S

pan

Diffusion Factor

Redesign

HTO300 S3

HTO300 EGV

APNASA S3

APNASA EGV

computational analysis and proposed redesign of the 3-stage axial compressor … · 2017. 5....

Documents